Implement reduced row echelon form function
This commit is contained in:
parent
f404bcbd6d
commit
6560411879
34
nalgebra-lapack/tests/linalg/rref.rs
Normal file
34
nalgebra-lapack/tests/linalg/rref.rs
Normal file
@ -0,0 +1,34 @@
|
||||
use na::rref;
|
||||
|
||||
use std::cmp;
|
||||
|
||||
use na::{DMatrix, DVector, Matrix4x3, Vector4};
|
||||
use nl::Cholesky;
|
||||
|
||||
use crate::proptest::*;
|
||||
use proptest::{prop_assert, proptest};
|
||||
|
||||
proptest! {
|
||||
#[test]
|
||||
pub fn rref_test() {
|
||||
let mat: Mat4 = Mat4::Identity();
|
||||
let res = rref(&mat);
|
||||
|
||||
assert_eq!(mat, res);
|
||||
|
||||
let m = Matrix3x4::<f64>::new(
|
||||
1.0, 2.0, -1.0, -4.0,
|
||||
2.0, 3.0, -1.0, -11.0,
|
||||
-2.0, 0.0, -3.0, 22.0,
|
||||
);
|
||||
|
||||
let expected = Matrix3x4::<f64>::new(
|
||||
1.0, 0.0, -1.0, -8.0,
|
||||
0.0, 1.0, 0.0, 1.0,
|
||||
0.0, 0.0, 1.0, -2.0,
|
||||
);
|
||||
|
||||
let res = rref(&mat);
|
||||
prop_assert!(relative_eq!(res, expected, epsilon = 1.0e-5));
|
||||
}
|
||||
}
|
@ -21,6 +21,7 @@ mod lu;
|
||||
mod permutation_sequence;
|
||||
mod pow;
|
||||
mod qr;
|
||||
mod rref;
|
||||
mod schur;
|
||||
mod solve;
|
||||
mod svd;
|
||||
|
64
src/linalg/rref.rs
Normal file
64
src/linalg/rref.rs
Normal file
@ -0,0 +1,64 @@
|
||||
//! Functions for computing the rref of a matrix.
|
||||
|
||||
use core::ops::{Mul, Sub, SubAssign};
|
||||
use simba::scalar::{ClosedMul, RealField};
|
||||
use std::ops::{DivAssign, MulAssign};
|
||||
|
||||
use crate::allocator::Allocator;
|
||||
use crate::base::dimension::Dim;
|
||||
use crate::base::{Const, DefaultAllocator, OMatrix, OVector, Scalar};
|
||||
|
||||
// Implementation of:
|
||||
// https://rosettacode.org/wiki/Reduced_row_echelon_form
|
||||
/// Compute the reduced row echelon form of a matrix.
|
||||
pub fn rref<T: RealField, D: Dim>(matrix: &OMatrix<T, D, D>) -> OMatrix<T, D, D>
|
||||
where
|
||||
DefaultAllocator: Allocator<T, D, D> + Allocator<T, D>,
|
||||
DefaultAllocator: Allocator<T, crate::base::dimension::Const<1>, D> + Allocator<T, D>,
|
||||
T: Scalar + ClosedMul,
|
||||
{
|
||||
let mut matrix = matrix.clone();
|
||||
let mut lead = 0;
|
||||
let row_count = matrix.nrows();
|
||||
let column_count = matrix.ncols();
|
||||
|
||||
for r in 0..row_count {
|
||||
if column_count <= lead {
|
||||
break;
|
||||
}
|
||||
|
||||
let mut i = r;
|
||||
|
||||
// Should this have an epsilon comparison instead?
|
||||
while matrix[(i, lead)] == crate::convert(0.0) {
|
||||
i += 1;
|
||||
if row_count == i {
|
||||
i = r;
|
||||
lead += 1;
|
||||
if column_count == lead {
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
matrix.swap_rows(i, r);
|
||||
|
||||
if matrix[(r, lead)] > T::default_epsilon() {
|
||||
let t = matrix[(r, lead)].clone();
|
||||
matrix.row_mut(r).div_assign(t);
|
||||
}
|
||||
|
||||
for i in 0..row_count {
|
||||
if i != r {
|
||||
let lv = -matrix[(i, lead)].clone();
|
||||
let r_row = matrix.row(r).clone() * lv;
|
||||
|
||||
matrix.row_mut(i).sub_assign(r_row);
|
||||
}
|
||||
}
|
||||
|
||||
lead += 1;
|
||||
}
|
||||
|
||||
matrix
|
||||
}
|
Loading…
Reference in New Issue
Block a user